As we approach the middle 1/3rd of the Tar Heels' basketball season, each year I like to try to get a feel for how the current team stands against historical UNC teams. My favorite standards for comparison are Points Per Possession for Offense and Defense. The following graph shows how each UNC team performed from 1987-2022 (since the national adoption of the 3-point shot). I'll discuss the methods below.
Current Team
After 10 games we see that the current team is exceptional at Offensive efficiency. In fact, only 5 teams have scored more easily than this team, and 4 of them went to the Final Four. Conversely, this is the second-worst defensive team we have put on the court. That stat has improved much in the last 5 games, however.
Dean Smith never wanted to compare teams, and the game of predicting outcomes can be dangerous. However, I have found that if one draws a diagonal line with a positive slope, we can see teams of relatively equal effectiveness. For example, the current team seems to be playing with a similar effectiveness to those teams of 2015, 2000, 1999, and 2011.
There is an interesting trend in this graph. Follow the teams from 2012 onward. You'll see a general trend to the right, showing that defensive performance in this program has deteriorated at a fairly constant rate. Defense is perhaps the best tell-tale of the players' respect for the coach. Progressively over the last 9 seasons, Roy Williams was motivating his players to play good defense less and less. This wears on a coach, so I wasn't terribly surprised when the coach announced his retirement. It's still quite early in Hubert Davis' career to draw any kind of conclusions about the meaning of the current team's poor defensive play, though.
When I report a statistic, I like to provide not only the 36-season average for that stat, but also the standard deviation. This give me a sense of how far from the average I should expect the statistic to be. Standard Deviation tells us a lot about how common a statistic is. About 2/3s of the time we'll be +/- 1 std dev from the average, 95% of the time we'll be +/- 2 std devs, and 99.7% of the time we'll be +/- 3 std devs.
UNC is shooting 40.9% from 3-point land thus far. Over the past 27 seasons we've shot 35.9% from 3 with a standard deviation of 2.6. My notation for this is: 35.9% (2.6). So:
Some other offensive stats are:
This is where the discussion of the rotation comes into play. We've seen small rotation teams play excellent basketball in the first half of the season. However the grind always catches up with these teams by February. My concern with this team is that with such a small rotation, players' legs will be the first to go when fatigue sets in, and shooting will follow. Since the entire offense is propped up with jump shooting, fatigue would ravage this team's ability to reach its potential.
Points Per Possession Computation
I use the Dean Smith method of computing possessions. Smith professed that a possession ends when the team loses control of the ball. The three situations where this happens are a turnover, a field goal attempt, or a trip to the FT line with 2 shots, 3 shots, or a 1-and-1 situation. If a team rebounds its miss, that begins a new possession. This is in stark contrast with the contemporary method of computing possessions. Today's method subtracts Offensive Rebounds from the total I compute, resulting in much higher points per possession figures.
The problem is rebounding. The contemporary method doesn't reflect the offense's performance as accurately because it is polluted with rebounding. Rebounding is like getting more arrows in archery, but it doesn't make you more accurate with the bow. My goal is to measure the accuracy of an offense.
How, then, is rebounding reflected with the Smith method? Because rebounds are not subtracted, each team will end up with a different number of possessions. That differential reflects the true rebounding from the game, according to Smith.
Additionally, Offensive Rebounds were not published for many of the games in the period of my study. Because this is a longitudinal study, using Smith's method works well for comparing this team to previous teams. If one is comparing UNC to other teams this season, a cross-sectional analysis, the contemporary method of computing possessions works fine.
I'll be updating this graph from time to time throughout the season. There is already been considerable movement for this team, and I'm sure that will continue.
- Better Offense is higher on the graph (larger Y) while better Defense is shown to the left (lower X). Therefore we want our teams be closer to the upper left corner.
- The current team is shown in red
- Final Four teams are shown in orange
- The green zone includes the teams that surpassed Dean Smith's goals of scoring over 0.95 points per possession while keeping the opponent below 0.85 points per possession
- The pink zone includes teams that neither scored over 0.95 pts/poss nor kept teams below 0.85 pts/poss.
Current Team
After 10 games we see that the current team is exceptional at Offensive efficiency. In fact, only 5 teams have scored more easily than this team, and 4 of them went to the Final Four. Conversely, this is the second-worst defensive team we have put on the court. That stat has improved much in the last 5 games, however.
Dean Smith never wanted to compare teams, and the game of predicting outcomes can be dangerous. However, I have found that if one draws a diagonal line with a positive slope, we can see teams of relatively equal effectiveness. For example, the current team seems to be playing with a similar effectiveness to those teams of 2015, 2000, 1999, and 2011.
There is an interesting trend in this graph. Follow the teams from 2012 onward. You'll see a general trend to the right, showing that defensive performance in this program has deteriorated at a fairly constant rate. Defense is perhaps the best tell-tale of the players' respect for the coach. Progressively over the last 9 seasons, Roy Williams was motivating his players to play good defense less and less. This wears on a coach, so I wasn't terribly surprised when the coach announced his retirement. It's still quite early in Hubert Davis' career to draw any kind of conclusions about the meaning of the current team's poor defensive play, though.
When I report a statistic, I like to provide not only the 36-season average for that stat, but also the standard deviation. This give me a sense of how far from the average I should expect the statistic to be. Standard Deviation tells us a lot about how common a statistic is. About 2/3s of the time we'll be +/- 1 std dev from the average, 95% of the time we'll be +/- 2 std devs, and 99.7% of the time we'll be +/- 3 std devs.
UNC is shooting 40.9% from 3-point land thus far. Over the past 27 seasons we've shot 35.9% from 3 with a standard deviation of 2.6. My notation for this is: 35.9% (2.6). So:
- 67% of the time UNC will shoot between 33.3% - 38.5%
- 95% of the time UNC will shoot between 30.7% - 41.1%
- 99.7% of the time UNC will shoot between 28.1% - 43.7%
Some other offensive stats are:
- Assists Per Possession: '22: 0.184. AVG: 0.199 (0.014)
- Pct Loss of Ball: '22: 0.153. AVG 1.60 (0.02)
- FT Attempts: '22: 19.5. AVG: 22.7 (2.6)
- Poss Differential: '22: 3.40. AVG: 2.00 (2.37)
This is where the discussion of the rotation comes into play. We've seen small rotation teams play excellent basketball in the first half of the season. However the grind always catches up with these teams by February. My concern with this team is that with such a small rotation, players' legs will be the first to go when fatigue sets in, and shooting will follow. Since the entire offense is propped up with jump shooting, fatigue would ravage this team's ability to reach its potential.
Points Per Possession Computation
I use the Dean Smith method of computing possessions. Smith professed that a possession ends when the team loses control of the ball. The three situations where this happens are a turnover, a field goal attempt, or a trip to the FT line with 2 shots, 3 shots, or a 1-and-1 situation. If a team rebounds its miss, that begins a new possession. This is in stark contrast with the contemporary method of computing possessions. Today's method subtracts Offensive Rebounds from the total I compute, resulting in much higher points per possession figures.
The problem is rebounding. The contemporary method doesn't reflect the offense's performance as accurately because it is polluted with rebounding. Rebounding is like getting more arrows in archery, but it doesn't make you more accurate with the bow. My goal is to measure the accuracy of an offense.
How, then, is rebounding reflected with the Smith method? Because rebounds are not subtracted, each team will end up with a different number of possessions. That differential reflects the true rebounding from the game, according to Smith.
Additionally, Offensive Rebounds were not published for many of the games in the period of my study. Because this is a longitudinal study, using Smith's method works well for comparing this team to previous teams. If one is comparing UNC to other teams this season, a cross-sectional analysis, the contemporary method of computing possessions works fine.
I'll be updating this graph from time to time throughout the season. There is already been considerable movement for this team, and I'm sure that will continue.
Last edited:
